Minutes of the AFE meeting, May 29, 2001 ----------------------------------------- John Anderson, Levan Babukhadia, Mrinmoy Bhattacharjee, Jerry Blazey, Fred Borcherding, Alan Bross, Satish Desai, George Ginther, Paul Grannis, Stefan Gruenendahl, Marvin Johnson, Daniel Mihalcea, Thomas Nunnemann, Abid Patwa, Harry Weerts, John Womersley, Mirek Zielinski, Vishnu Zutshi, 1. AFE Production. -------------- The 54 extra bare boards have arrived. 20 were found to have countersunk holes on the wrong side, and there are concerns that these may weaken the backing bar connection. George G. will test one of these. Negotiations are underway with the manufacturer to repair or replace these boards. Parts for the Rev3CE (extra) are mostly in hand, but will probably take another two weeks to accumulate. The bid for assembly has not been released by Fermilab. Sanmina now predicts production of the RH board stuffing for Monday, June 4. The added delay was just due to internal scheduling. John Anderson and Johnny Green will arrive at Sanmina on Monday to oversee the stuffing and discuss with the production crew. They hope to bring back at least one board for us to start testing here. John thinks the LH board assembly will be roughly 2 weeks later, but will discuss with them directly. Some questions were raised on whether we should hold the LH boards until tests are done on RH boards. John was not sure this could help, as the issues are primarily things like part placement and orientation that can go wrong or right independently of testing. There may also be enough time to complete LAB3 tests in the 2 week interval between the LHB and RHBs. 2. Testing at FNAL --------------- Fred B. reported that Bob Angstadt is doing the test software for the DAB3 test, and is making good progress. Mike Matulik is progressing well with the AFE test stand hardware. However, there are at present still problems with noise in the charge injection mode. We need to maintain the pressure to get the test stand up and running. Daniel M. reported that the SiDet test procedures and manpower issues look to be under control. One test stand will be complete by the end of the week. We think there are at present about 5 testers for DAB3 signed up, with promissory notes from several other groups. Rick Jesik should provide an accounting of test personnel next Monday on return from London, and management should proceed to get the needed complement of testers. Our previous plan was to man two shifts per day, with 4 hour shifts. Training should begin in about a week. Marvin emphasized that getting 20 good boards for testing and commissioning is very important. We should initially move to get this first complement, before trying to fix those with problems. 3. MCM availability ---------------- Fred B. reported on the inventory of MCMs. His spreadsheet is referenced on the web page. We initially bought 2214 MCMs from Meltronix (300 additional substrates are banked). Of these, 1845 worked. We expect that relatively straightforward fixes would recover another 185, giving us 2030 working MCMs. We need a total of 1680, exclusive of the FPS MCMs (80 RevA boards, 1168 RevC-S(anmina) and 432 RevC-E(xtra) ) MCMs. With 36 FPS AFE8p (read AFE8prime) boards, the total comes to 1968. With 36 FPS AFE12s, the need would total 2112. Clearly MCM availability is an issue. 4. AFE8p for FPS ------------- Paul G. summarized the issues for choice of AFE8p vs. AFE12 for FPS. The issues are (1) saturation of SVX readout, (2) relative calibration with MIPs, (3) loss of dual trigger threshold, (4) handling crossover channels, (5) MCM availability, and (6) time to complete. We would like to maintain a full scale on the SVX for standard running of about 70 MIPs (the mean deposit in shower layers for E (not ET) = 90 GeV electrons. We would like to be able to obtain roughly 10 counts/MIP for stable MIP calibration in special runs. A spreadsheet was shown (see afe web page) that accomodates this with a 10 pF input capacitor to the FPS SIFTs. There is not much margin however. Loss of dual range triggering can of course be a problem; we would try to run with as low threshold as possible to retain low ET physics. We do not expect to need preshower for W/Z physics at less than full luminosity (the calorimeter and quadrant matching do a lot). The cross over channels (16 shower strips migrate to Modules 2&7) and 8 shower layer strips migrate from Module 3(6) to Module 4(5). For the FPS as cabled, these are the 16 strips closest to the beam in both orientations of FPS sectors. We expect the AFE8p to handle the distribution of appropriate bias voltages to the VLPCs. However, the AFE12 would have returned the cross over channels to the appropriate MCM (with all 72 channels used) and AFE8p does not. This means that the cross over channels which use VLPCs from different batches may have characteristics (gain, threshold) different from other channels on the same MCM. Moreover, we would want to have small input caps on the shower strips of Modules 2&7 while the same MCM MIP strips want large input capacitance. John A. thought putting different caps would not give a problem, but this has not been tested. Some differences of shower strips on shower MCMs 3&6 might be accomodated with modified input cap charge splits. It seems that some special handling of the cross over channels will be needed -- each individual sector of FPS will likely be somewhat different. MCM availability to serve AFE12s is very tight as mentioned above. The time to complete AFE8p was estimated as 4-6 months by John Anderson last week. AFE12 completion is estimated at 10-12 months. Paul made a pitch that we look at performance of FPS (and CPS) with AFE8 using beam before stuffing the final input capacitors. There is little margin on the charge split value for either. Several noted that the relevant information could be obtained soon using stereo boards, and the known comparison parameters. The time required to re-layout the AFE8 for AFE8p vs. simple modification to the existing AFE board was discussed, without clear understanding being reached. +--------------------------------------------------------------------+ | The group recommended that AFE8p be chosen for FPS operation. | | Details of using real AFE8 boards and adding bias lines with wires | | and Xacto knife vs. a re-layout of the AFE8 board was left for | | discussion in two weeks, based on the advice from the engineers. | +--------------------------------------------------------------------+ 5. FPS saturation -------------- Vishnu presented results on correcting the calorimeter energy for preshower deposits. He had not finished the FPS study, so focussed on CPS. He showed that moderate saturation (~50 MIP saturation) did not materially affect the calorimeter resolution. The effect of saturation is to make a more linear correspondence between CCEM1 and CPS. There was some puzzlement about how a linear correlation could improve resolution. Vishnu pointed out that the correction is actually dependent on energy, and that there could be differences in correcting photons and electrons. Nevertheless, for CPS, it seems that we can live with the level of saturation that we will encounter using AFE8. The results for FPS will be posted when available. 6. MCM replacement --------------- Marvin reported on work on the MCM replacement using FADCs and SVX4. There was a previous error (by factor of 2) in the packing density of signals needed for dual range thresholds. The solution for this is still under study. However the prospect exists that with the available wires in place for Vbias on AFE8p and the LVDS output capability, a simple MCM replacement can be quite attractive. The modification for 132 ns operation would be needed for all channels; some would also need the Vbias changes. Work continues on design. Johnny Green is working on a cost estimate for the MCM replacement. Meltronix has been asked to give an advisory quote for SIFT replacement on the existing MCMs, so that we have a comparison.